Modeling sub-band correlation for noise-robust speech recognition.
James McAuleyJi MingPhilip HannaDarryl StewartPublished in: ICASSP (1) (2004)
Keyphrases
- speech recognition
- noisy environments
- subband
- low frequency
- automatic speech recognition
- hidden markov models
- language model
- speech recognition technology
- speech synthesis
- multiresolution
- speech signal
- speaker identification
- background noise
- wavelet transform
- high frequency
- discrete wavelet transform
- speech recognizer
- pattern recognition
- filter bank
- image compression
- wavelet decomposition
- high pass
- frequency domain
- speech enhancement
- noisy speech
- additive noise
- noise reduction
- low pass
- bit rate
- feature vectors
- speech recognition systems
- speaker independent
- wavelet coefficients
- frequency band
- image data
- high quality